ORIGINAL ARTICLE

Predicting the intervention threshold for initiating osteoporosistreatment among postmenopausal women in China:a cost-effectiveness analysis based on real-world data

L. Cui1 & T. He2 & Y. Jiang1& M. Li1 & O. Wang1

& R. Jiajue1 & Y. Chi1 & Q. Xu3& X. Xing1

& W. Xia1

Received: 13 February 2019 /Accepted: 18 September 2019 /Published online: 21 November 2019#

AbstractSummary This study built a micro-simulation Markov model to determine the treatment threshold of osteoporosis in postmen-opausal women in Mainland China. Treatment with zoledronate is cost-effective when FRAX-based (Fracture risk assessmenttool) fracture probability is over 7%.Introduction The purpose of this study is to estimate FRAX-based fracture probabilities in Mainland China using real-worlddata, at which intervention could be cost-effective.Methods We developed a micro-simulation Markov model to capture osteoporosis states and relevant morbidities including hipfracture, vertebral fracture, and wrist fracture. Baseline characteristics including incidences of osteoporosis and distribution ofrisk factors were derived from the Peking Vertebral Fracture study, the largest prospective cohort study of postmenopausalwomen in Mainland China. We projected incidences of fractures and deaths by age groups under two treatment scenarios: 1)no treatment, and 2) zoledronate. We also projected total quality-adjusted life-years (QALY) and total costs including fracturemanagement and osteoporosis drugs for cost-effectiveness analysis. Cost-effective intervention thresholds were calculated basedon the Chinese FRAX model.Results Treatment with zoledronate was cost-effective when the 10-year probability of major osteoporotic fracture based onFRAXwas above 7%. The FRAX threshold increased by age from 51 to 65 years old, and decreased in elder age groups, rangingfrom 4% to 9%.Conclusions Using real-world data, our model indicated that widespread use of zoledronate was of both clinical and economicbenefit among Chinese postmenopausal women. Using a FRAX-based intervention threshold of 7% with zoledronate shouldpermit cost-effective access to therapy to patients and contribute to reducing the disease burden of osteoporosis in MainlandChina.

Keywords Chinese . cost-effectiveness . FRAX .Markov . osteoporosis . zoledronate

Background

Osteoporosis is a major public health concern in MainlandChina. It was estimated that there were up to 70 million oste-oporotic patients, and over 200 million patients withosteopenia in China in 2006 [1]. Osteoporotic fracture is themost severe complication of osteoporosis and results in hos-pitalization, disability, and mortality. The prevalence of verte-bral fracture in postmenopausal Chinese women increasedfrom 13.4% at ages 50–59 years, to 58.1% at age 80 yearsor older [2]. The overall major osteoporotic fracture inChinese population was estimated to be 2.69 million casesin 2015 [3], indicating an enormous disease burden of osteo-porotic fractures in China.

Electronic supplementary material The online version of this article(https://doi.org/10.1007/s00198-019-05173-6) contains supplementarymaterial, which is available to authorized users.

* W. Xiaxiaweibo8301@163.com

1 Department of Endocrinology, Key Laboratory of Endocrinology,Ministry of Health, Peking UnionMedical College Hospital, ChineseAcademy of Medical Sciences, Beijing 100730, China

2 Department of Hematology, PekingUnionMedical College Hospital,Chinese Academy of Medical Sciences, Beijing 100730, China

3 Department of Orthopedics, Beijing Jishuitan Hospital,Beijing 100035, China

Osteoporosis International (2020) 31:307–316https://doi.org/10.1007/s00198-019-05173-6

The Author(s) 2019

Osteoporosis treatment beforehand exhibits efficacy infracture risk reduction. Of all treatment regimens, bisphospho-nate is the most commonly used drug and was predicted to becost-effective in other models [4]. In one of our previous stud-ies (unpublished), alendronate was dominated by zoledronatein cost-effectiveness analysis, suggesting zoledronate a betteroption for treating Chinese postmenopausal women. InMainland China, there is no universally accepted policy fortreating patients at high risk, and only a minority of osteopo-rotic people receive proper treatment due to lack of evidence-based guidelines and recommendations [5]. Fracture risk as-sessment tool (FRAX) is a popular tool worldwide, used topredict 10-year probability of osteoporotic fracture based onclinical risk factors and femoral neck bone mineral density(BMD) [6]. Several cost-effectiveness analyses based onWestern settings suggest that osteoporotic treatment shouldbe initiated when the 10-year major fracture probability, basedon FRAX, is over 7-15% [7–10]. Previous studies on estimat-ing the FRAX threshold for the Chinese population were onlylimited to clinical observations or expert consensus [11–14].

We introduced real-world data from the Peking VertebralFracture (PK-VF) study into modeling to improve the robust-ness of cost-effectiveness analysis. The PK-VF study is cur-rently the largest prospective cohort study of postmenopausalosteoporotic women inMainland China, which involved 1100women and follow-up from 2008 to 2013, to capture baselineepidemiological data of osteoporosis and osteoporotic verte-bra fracture [2, 15–17]. The epidemiological data of the cross-sectional survey in 2008 has been published, including prev-alence of osteoporosis and osteoporotic vertebra fracture andthe distribution of risk factors of osteoporosis in the popula-tion [2].

In this study, we built a micro-simulation Markov modelbased on the PK-VF study. We aim to estimate the interven-tion threshold of FRAX in Mainland China, at which treat-ment with zoledronate could be cost-effective.

Methods

Model approach

We developed an individual-based micro-simulation Markovmodel in NetLogo 5.3.1 (Evanston, IL) to evaluate cost-effectiveness of osteoporotic treatment with zoledronate underdifferent initiation thresholds. Subjects in the model were sim-ulated on the following aspects: age, BMI, osteoporosis status,fracture status, risk factors of fracture (including previous frac-ture, parent fracture, smoking, alcohol, use of glucocorticoid,and the history of rheumatoid arthritis and secondary osteopo-rosis), and medical costs (including drug costs and fracturemanagement costs) (Fig. 1a). Subjects in the model couldtransit between different states, including states of non-

fracture, fracture, and postfracture, as shown by arrows (Fig.1b). We simulated 20,000 people representing Chinese post-menopausal women as a closed cohort with a lifelong timehorizon. A discount rate was set as 3% per year for both costsand effectiveness. Cost-effectiveness was considered as ICERless than 3 times of per capita GNP, which was $20,000/QALY for 2018 China.

Baseline population from real-world data

We incorporated characteristics from the PK-VF study into themodel with a stochastic sampling method for baseline popu-lation setup. By initial setting up the population, individuals’characteristics were distributed according to the PK-VF study,including eight risk factors related to osteoporotic fracture inFRAX that is body mass index (BMI), previous fracture, par-ent fracture hip, current smoking, glucocorticoid use, rheuma-toid arthritis, secondary osteoporosis, and alcohol use. Foreach age group, we randomly selected a number of individualsof the same age, together with their osteoporotic status, ac-companied risk factors and treatment histories, from PK-VFparticipants. The percentage of each clinical risk factor basedon PK-VF cohort was listed in Supplementary Table 1. Asmost of the risk factors (e.g., parent fracture hip, currentsmoking, glucocorticoid use, rheumatoid arthritis, BMI, sec-ondary osteoporosis, and alcohol use) may have sustainedimpact on the osteoporosis-related fractures, we assumed thatthe risk factors of individuals simulated in the model remainunchanged except for previous fractures, which is updated ifosteoporotic fractures occurred. For example, we assumed theeffect of smoking on the increased risk of osteoporosis re-mains sustained no matter the individual stopped smoking ornot.

Fracture risks and mortality

Our model began with women in the no-fracture state andsimulated yearly transition till death. Osteoporosis developedwith a yearly probability by age derived from the age-specificincidence of osteoporosis from PK-VF study (Table 1). Theincidence of vertebral fracture, hip fracture, and wrist fracturewere from pooled data of the Hong Kong Osteoporosis Study,the Hefei osteoporosis project, and a Norwegian study (mul-tiplied by 0.72 to adjust for the Chinese population) [18–20](Table 1). Osteoporotic subjects were more likely to experi-ence fractures, with the relative risks published by Meltonet al. [21]. The risk of a subsequent fracture increased whena prior fracture occurred [22]. Age distribution and naturalmortality rate were from the 2012 National Sample Surveyon Population Changes [33]. Standardized mortality ratios(SMR) of vertebral fracture, hip fracture, and wrist fracturewere set as 1.82, 2.43, and 1.42 respectively [23] (Table 1).

308 Osteoporos Int (2020) 31:307–316

Treatment strategies

We incorporated the treatment of zoledronate into themodel, dosed as 5 mg a year, combined with calciumand vitamin D3 supplement (calcium 600 mg + vitaminD3 125IU per tablet), dosed once a day. The effect ofzoledronate on reducing fracture risks was taken fromclinical trials [26, 34, 35].

We calculated the 10-year major fracture probability(FRAX score) for each subject based on FRAX equation [6],

which incorporated covariates including age and risk factors.Those with an estimated FRAX score higher than the thresh-old set were eligible for treatment. The antiosteoporosis treat-ment was initiated for subjects when FRAX score was higherthan the threshold, or when fracture occurred in the subject(Fig. 1c). Once initiated, those subjects would be treated forfive years under each of the two treatment scenarios: (1) notreatment or (2) zoledronate. The remaining effect was as-sumed to decrease linearly in the next 5 years after the treat-ment was over [36].

Fig. 1 Schematic of the modelstructure. a Subjects in the modelwere simulated on the aspects ofdemographic feature, healthfeature, risk factor feature, andtreatment feature. b Subjectscould transit between differentstates, including states of non-fracture, fracture, andpostfracture, as shown by arrows.Multiple fractures were allowedto happen in the same subjectsduring the lifetime. All stateswould also be absorbed into deathstate. c The anti-osteoporosistreatment would be initiated tosubjects when FRAX score washigher than the threshold, or whenfracture happened in the subject.Once initiated, we treated thosesubjects for five years under eachof the two treatment scenarios,including (1) no treatment and (2)zoledronate. Subjects might drop-off during the 5-year treatment,and the remaining effect woulddecrease linearly in the next 5-year posttreatment

Osteoporos Int (2020) 31:307–316 309

Imperfect adherence and persistence may affect the cost-effectiveness of the treatment and were thus incorporated inthe model (Table 1) [27, 37]. Adherence was defined as the

extent to which the subject acted in accordance with the pre-scribed drug, with respect to dosage and timing [30]. We setthe adherence of zoledronate to be 1, as zoledronate was

Table 1 Key parameters derived from Peking-Vertebral Fracture (PK-VF) study and from literatures

Parameters Distribution

Prevalence of osteoporosis (%)(PK-VF study)

5.26 (50–54 years), 13.1 (55–59 years), 19.6 (60–64 years), 26.4 (65–69 years), 24.5 (70–74 years), 27.2 (75–79years), 30.1 (80–84 years), 35.0 (85 + years)

Incidence of osteoporosis (%)(PK-VF study)

1.91 (50–54 years), 2.80 (55–59 years), 1.80 (60–64 years), 1.04 (65–69 years), 2.10 (70–74 years), 2.22 (75–79years), 3.96 (80 + years)

Fracture incidence (‰)Vertebral fracture [18] 2.19 (50–54 years), 3.13 (55–59 years),

5.16 (60–64 years), 5.64 (65–69 years),8.74 (70–74 years), 12.05 (75–79 years),21.19 (80–84 years), 26.89 (85–89 years),27.10 (90 + years)

-

Hip fracture [19] 0.33 (50–54 years), 0.46 (55–59 years),0.54 (60–64 years), 0.96 (65–69 years),2.33 (70–74 years), 4.08 (75–79 years),6.44 (80–84 years), 6.59 (85–89 years),8.67 (90 + years)

-

Wrist fracture [20]a 4.76 (50–54 years), 7.32 (55–59 years),11.16 (60–64 years), 12.95 (65–69 years),13.17 (70–74 years), 13.87 (75–79 years),15.01 (80–84 years), 15.10 (85–89 years),13.97 (90 + years)

-

Osteoporosis attribution probabilities by fracture type [21]Vertebral fracture 0.75 (range 0.40–0.80) (45–64 years), 0.75 (range 0.50–0.90) (65–84 years), 0.95 (range 0.60–0.95) (85 + years) TriangularHip fracture 0.75 (range 0.20–0.85) (45–64 years), 0.85 (range 0.50–0.95) (65–84 years), 0.95 (range 0.60–0.95) (85 + years) TriangularWrist fracture 0.60 (range 0.10–0.70) (45–64 years), 0.70 (range 0.35–0.80) (65–84 years), 0.70 (range 0.55–0.90) (85 + years) TriangularRelative risk of subsequent fracture following a prior fracture [22]Vertebral fracture 2.0 (range 1.6–2.4) GammaHip fracture 2.0 (range 1.9–2.2) GammaWrist fracture 1.9 (range 1.6–2.2) GammaStandardized mortality ratios (SMR) [23]Vertebral fracture 1.82 (95% CI 1.52–2.17) GammaHip fracture 2.43 (95% CI 2.02–2.93) GammaWrist fracture 1.42 (95% CI 1.19–1.70) GammaQuality-adjusted life year (QALY)By age [24] 0.772 (55–59 years), 0.728 (range 0.582–0.874) (60–64 years), 0.702 (range 0.562–0.842) (65–69 years), 0.685

(range 0.548–0.822) (70–74 years), 0.669 (range 0.535–0.803) (75–79 years), 0.655 (range 0.524–0.786)(80–84 years), 0.643 (range 0.514–0.77) (85 + years)

Triangular

Vertebral fracture [25] 0.724 (range 0.667–0.779) (1st year), 0.868 (range 0.827–0.922) (subsequent years) BetaHip fracture [25] 0.776 (range 0.720–0.844) (1st year), 0.855 (range 0.800–0.909) (subsequent years) BetaWrist fracture [25] 1.000 (range 0.960–1.000) (1st year), 1.000 (range 0.930–1.000) (subsequent years) TriangularTreatment of zoledronate related parametersRelative risk of fracture ofzoledronate [26]

0.23 (range 0.14–0.37) (vertebral fracture), 0.59 (range 0.42–0.83) (hip fracture), 0.75 (range 0.64–0.87) (wristfracture)

Lognormal

Adherence of zoledronateb 1Persistence of zoledronatec [27] 1 (1st year), 0.748 (subsequent years)COST (2019 US dollar)Vertebral fracture 1st year [32]d 3411Hip fracture 1st year [32]d 4514Wrist fracture 1st year [32]d 1383Zoledronate + Ca + Vit D [28] 492Annual discount rate [29] 0.03 (range 0–0.05) -

a Incidence of wrist fracture was from Norwegian population, and multiplied by 0.72 to adjust for Chinese population [20]b Adherence referred to the extent to which the subject acted in accordance with the prescribed drug, with respect to dosage and timing [30]. Theadherence of zoledronate was hypothesized to be 1, as zoledronate was injected once a yearc Persistence referred to the duration from initiation until discontinuation of the treatment with a certain drug [30]d Costs of vertebral, hip, and wrist fracture referred to all direct medical costs and direct non-medical costs reported by Qu et al. in 2012 [31]. Directmedical costs included costs of outpatient and inpatient care, examination, medication, rehabilitation and physical therapy after fracture; and direct non-medical costs included traffic fee, preventive care foods, and specific equipment. All costs (collected in 2012) were converted to 2019 RMB by aninflation rate of 1.1466 [32]

310 Osteoporos Int (2020) 31:307–316

injected once a year. Subjects in the model may also drop outduring the 5-year treatment, and persistence was defined as theduration from initiation until discontinuation of the treatmentwith a certain drug [30].

Cost assumptions

Costs of vertebral, hip, and wrist fracture referred to all directmedical costs and direct nonmedical costs reported by Quet al. in 2012 [31]. Direct medical costs included costs ofoutpatient and inpatient care, examination, medication, reha-bilitation, and physical therapy after fracture; direct non-medical costs included traffic fee, preventive care foods, andspecific equipment. All fracture costs (data collected in 2012)were converted to 2018 RMB by a total inflation rate of1.1466 reported by inflation.eu [32]. Costs of zoledronate(100 ml: 5 mg per year) and calcium + vitamin D supplement(calcium 600 mg + vitamin D3 125IU per tablet per day) wasobtained from the Chinese market drug price database (accessdate Feb 07, 2019) [28]. RMB was converted to US dollars atthe exchange rate of 6.7426 (Feb 07, 2019).

Utilities

Quality-adjusted life year (QALY) is a measurement of bothquality and quantity of life lived, varying from 0 to 1, where 0referring to death and 1 referring to perfect health. The age-specific QALY for the general population (Table 1) was basedon the National Health Services Survey 2008 in China [24].The QALYof different fracture types in the general populationand osteoporotic population (Table 1) were obtained from ameta-analysis [25].

Presentation of results

We projected QALY, as a measurement of health and totalcosts of both fracture management and osteoporosis drugsfor the cost-effectiveness analysis. The incremental cost-effectiveness ratio (ICER) was calculated as (cost A–cost B)/(QALYA–QALY B). The willingness to pay (WTP) was as-sumed to be 3 times of per capita GNP, which was $20,000/QALY for 2018 China.

Sensitivity analysis

To test the robustness of the model to uncertainties of inputdata, one-way sensitivity analysis comparing the treatment ofzoledronate to no treatment was performed on key model pa-rameters, including fracture incidence, incidence of osteopo-rosis, standard mortality ratio, relative risk of subsequent frac-ture following a prior fracture, QALY by age, QALY by frac-ture types, relative risk of fracture with treatment, persistence,treatment year, adherence, drug cost, and annual discount. The

upper and lower limits of each key parameter are listed inTable 1 and Appendix.

Results

Validation

Face validation was done by consulting experts on osteoporo-sis prevention and treatment. Internal validation was per-formed on comparison of annual fracture incidences frommodel prediction to those from input data (SupplementaryTable 2). The projected incidences were all within ± 5% ofthe reference value. External validation was performed oncomparison of life expectancy, prevalence of osteoporosis,and lifetime fracture risk predicted from model to those fromliterature (Supplementary Table 3) [38–40]. Cross-model val-idation was not performed, as no modeling work has beendone on estimating the intervention threshold of Chinesemainland population.

Intervention FRAX threshold estimation

Comparing the no-treatment scenario to treatment withzoledronate, the total costs increased from $1407.4 to$1822.9 per person, and the corresponding QALYs increasedfrom 9.232 to 9.269. The ICER was $26637.3, $22129.7,$20338.4, $19285.0, $18181.0, $16680.2, $15047.7, and$14447.7 per QALY at FRAX threshold 0.02, 0.06, 0.07,0.08, 0.09, 0.1, 0.5, and 1, respectively. As Fig. 2 indicated,the overall treatment threshold was 7% for treatment withzoledronate to be cost-effective in comparison to no treatment.

Furthermore, we conducted cost-effectiveness analysis byage groups. As Fig. 3 and Supplementary Table 4 indicated,FRAX thresholds in different age groups ranged from 0.040 to0.090. The threshold increased by age from 51 to 65 years old,and decreased in elder age groups.

We also performed a subgroup cost-effectiveness analysison subjects all with previous fracture history (SupplementaryTable 5, Fig. 4). By comparing Supplementary Table 5 toSupplementary Table 4, treatment with zoledronate was foundto be cost-saving in the youngest and oldest age groups withprevious fracture, while in other age groups, the treatmentthreshold based on FRAX was similar between people withand without previous fracture.

Sensitivity analysis

We performed sensitivity analysis on key model parameters,comparing the treatment of zoledronate to no treatment(Supplementary Table 6). As shown in the tornado plot (Fig.5), the drug cost of zoledronate, incidence of vertebral fracturein osteoporosis, and QALY by age and persistence were the

Osteoporos Int (2020) 31:307–316 311

most sensitive parameters, but ICERs for all parameters re-main below $23,000/QALY.

Discussion

In this study, we built an individual-based Markov model ofosteoporosis simulating postmenopausal women in MainlandChina, based on real-world data from the PK-VF study. Wepredicted the intervention threshold of FRAX in MainlandChina at which treatment with zoledronate could be cost-effective and calculated the variation of the threshold in dif-ferent age groups.

In Mainland China, there is no accepted criterion for initi-ating the treatment of osteoporosis. Current expert consensus

in China is that anti-osteoporosis treatment should begin inany of the following conditions: (1) the occurrence of verte-bral fragile fracture or hip fragile fracture; (2) BMD of femoralneck, whole hip, lumbar vertebrae, or distal radius < − 2.5; (3)− 2.5 < BMD < − 1 plus (1) occurrence of the proximalhumerus, distal forearm, or pelvis fragile fracture or (2) 10-year hip fracture probability > 3% or 10-year major fractureprobability > 20% predicted by FRAX [41]. However, therehas been no cost-effectiveness analysis to determine aChinese-specific threshold to initiate osteoporosis treatmentbased on FRAX. Thus, building a cost-effectiveness modelthat predicts FRAX thresholds for the Chinese population isboth important and urgent.

Several cost-effectiveness analyses have been performedon predicting the disease burden of osteoporosis in postmen-opausal women in Mainland China [3, 42–45], but none ofthem were based on real-world data of osteoporotic cohorts inthe Chinese population. This study, however, was based onour previous PekingVertebral Fracture (PK-VF) [2, 15–17], inwhich 1100 Chinese postmenopausal women were followedup from 2008 to 2013, and data were collected on the inci-dence and prevalence of osteoporosis and osteoporotic verte-bra fracture, and the distribution of risk factors of osteoporosisin the population [2]. Introducing real-world data into model-ing would significantly improve the simulation.

The determination of cost-effectiveness FRAX interven-tion threshold should be country-specific, considering thelarge variation of epidemiology and economic conditionsacross countries. For example, osteoporosis treatment waspredicted to be cost-effective when 10-year major fractureexceeded 7% in UK [7], 15% in Switzerland [8], 10% inGreece [9], 8.8% in Portugal [10], and when 10-year hip frac-ture exceeded 3% in USA [46]. There have been a few studies

Fig. 3 Heatmap of incrementalcost-effectiveness ratio (ICER)for zoledronate by age-group andby FRAX threshold. Red linesindicated the willingness to pay(WTP) as $20,000/QALY

Fig. 2 FRAX threshold for zoledronate. Treatment with zoledronate forpatients with 10-year major fracture probability by FRAX over 0.07 wascost-effective in comparison with no treatment

312 Osteoporos Int (2020) 31:307–316

estimating the intervening threshold of osteoporosis forChinese population based on FRAX, but they were only lim-ited to clinical observations or expert consensus without theevaluation of the economic and disease burden of osteoporosis[11–14].

No cost-effectiveness analysis has been done on estimatingFRAX threshold in China so far. One of the reasons was thatreal-world data of the distribution of clinical risk factors werenot available in China before [45]. There were eight clinicalrisk factors in FRAXmodel to determine a 10-year probabilityof major fracture and hip fracture of an individual, including

age, sex, BMI, previous fracture, parent fracture hip, currentsmoking, glucocorticoid use, rheumatoid arthritis, secondaryosteoporosis, and alcohol use (with or without the addition offemoral neck BMD). As risk factors of osteoporosis werealways not independent of each other, it would lead to poten-tial bias if we were to bring each risk factor separately into themodel. For example, current smokers are more likely to be aregular alcohol drinker. To solve this issue, we generated asampling table of risk factors based on our PK-VF cohort,and randomly sampled a group of risk factors as a whole fromthe sampling table for each subject simulated in the model.

Fig. 5 Tornado plot of one-waysensitivity analysis on key modelparameters, comparing treatmentwith zoledronate to no-treatmentscenario. The upper and lowerlimits of all key model parametersrun for the sensitivity analysiswere listed in Table 1. ICER, in-cremental cost-effectiveness ratio;RR, relative risk; QALY, quality-adjusted life year

Fig. 4 Heatmap of incremental cost-effectiveness ratio (ICER) for zoledronate in subjects with previous fracture by age-group and by FRAX threshold.Red lines indicated the willingness to pay (WTP) as $20,000/QALY

Osteoporos Int (2020) 31:307–316 313

There are various anti-osteoporosis drugs among whichb i sphosphona t e i s t he mos t w ide l y u s ed one .Bisphosphonate functions in inhibiting the bone reabsorptionprocess of osteoclasts. Alendronate, an oral bisphosphonate, isdosed once per week, and zoledronate is dosed once a yearintravenously. Alendronate and zoledronate reduced the riskof vertebral fracture by 48% and 70% respectively in clinicaltrials [26, 47], and in one of our previous studies (unpub-lished), alendronate was dominated by zoledronate in cost-effectiveness analysis, suggesting zoledronate a better optionfor treating Chinese postmenopausal women.

As indicated in Fig. 2, FRAX threshold for the wholeChinese population was predicted to be 7%. FRAX thresholdsdiffered in different age groups (Fig. 3 and SupplementaryTable 4), as the number first increased in age groups youngerthan 65 years, and decreased in elder age groups. The reasonbehind this phenomenon is the bidirectional effect of aging onthe resultant gain of QALYs, and thus on the ICERs, in atreatment-as-prevention mode, (1) early intervention withlow FRAX threshold among younger patients, who have lon-ger expected lifetime, resulted in more cumulative QALYsgained. (2) Intensive intervention with low FRAX thresholdamong older patients, who have higher fracture risk, resultedin a marked reduction of osteoporotic fractures and relateddeaths and thus in more QALYs gained. Therefore, we sug-gested additional resources and attentions shift to the middleaged or advanced aged.

In the external validation table (Supplementary Table 3), agap was found between the projected life expectancy (16.49)and the reference life expectancy (16.9) in the age group of65–69 years. We think the reason for the gap was that we usedthe overall death rate by age from a national survey [33], andapplied osteoporosis, fractures, and deaths due to fractures tothe simulated population additionally. Therefore, the simulat-ed population presented a slightly higher death rate and thus ashorter life expectancy, especially among the elders. However,such an overestimated death rate would result in anunderestimated benefit of anti-osteoporosis treatment, becauseit relies on long-term follow-up to observe the benefit of treat-ment of osteoporosis in preventing fractures, and overestima-tion of the death rate tends to underestimate the preventivegain of QALYs and the cost-effectiveness of treatment.Currently, treatment of osteoporosis in clinical practice wasrather insufficient. With such a conserved estimation, we putforward a FRAX-based intervention threshold as 7%.

There are certain limitations to this study. First, not all dataneeded for modeling were available for the Chinese popula-tion. For example, the incidence of wrist fracture was adoptedfrom the Norwegian population, and multiplied by 0.72 toadjust to the Chinese population [20]. And we generally strat-ified transition-probabilities to osteoporotic status fromhealthy status by age, regardless of those listed risk factorssuch as BMI, parent fracture, current smoking. Since only

67 new cases of osteoporosis were observed in the 5-yearfollow-up of PK-VF study, the impact of these covariates onthe incidences was not available. And for the same reason, wetook only age, previous fractures, treatment, and osteoporoticstatus as impact factors on transition probabilities of fracturesbut did not considered the impact of all listed risk factorsabove. Second, our real-world data were collected in Beijingpopulation, which may not represent the complete picture ofthe Chinese population. Our group is now performing an on-going national-wide epidemiological study on osteoporosiswith over 20,000 people recruited, and will attempt to incor-porate these data into modeling in the future. Third, side ef-fects of zoledronate, such as osteonecrosis of the jaw andatypical femoral fracture, were not considered in the model.The reason is that the health state utility of these side effectswas unclear for Chinese population, and these effects occurrarely at the dose of treating osteoporosis. Fourth, one-waysensitivity analysis was performed in the study, which maynot be as robust as probabilistic sensitivity analysis.However, we found ICERs tested on all variables in the cur-rent sensitivity analysis fell below $23,000, suggesting theresults were not sensitive to the uncertainty of those input data.Fifth, we used visual analogue scale (VAS)-based QALY byfracture types as the input data of the model. Though a tradetime-off (TTO)-based QALYmodel was available for Chinesepopulation [48], we were unable to use it to calculate QALYby fracture types, as we did not collect the information in 5dimensions (i.e., mobility, self-care, usual activities,pain/discomfort, and anxiety/depression) and 3 levels (i.e., 1= no problems, 2 = some/moderate problems, and 3 = extremeproblems) for each individual in PK-VF study. As an alterna-tive, we used the VAS-based QALY data by fracture types inthe model [25], which has also been used in other previouscost-effectiveness analysis [43, 45].

Conclusions

This study is the first cost-effectiveness analysis simulatingosteoporosis and osteoporotic fractures in Chinese postmeno-pausal women based on real-world data. The study indicatedthat widespread use of zoledronate was of clinical and eco-nomic benefit to Chinese postmenopausal women with a 10-year major fracture probability by FRAX over 7%.

Funding information This study was supported by grants from ChinaPostdoctoral Science Foundation Grant (2018M631396), NationalNatural Science Foundation of China (No. 81070687, 81170805 and81670714), Beijing Natural Science Foundation (No. 7121012),National Key Program of Clinical Science (WBYZ2011-873) andCollaborative Innovation Team Project of Medical and Health Scienceand Technology Innovation Project of Chinese Academy of MedicalSciences (No. 2016-I2M-3-003).

314 Osteoporos Int (2020) 31:307–316

Osteoporos Int (2020) 31:307–316 315

Compliance with ethical standards

Conflicts of interest None.

Appendix: Checklists of economic evaluationson osteoporosis, based on ESCEO and IOFguideline, and CHEERS guideline [49, 50]

Open Access This article is distributed under the terms of the CreativeCommons Attribution-NonCommercial 4.0 International License (http://creativecommons.org/licenses/by-nc/4.0/), which permits anynoncommercial use, distribution, and reproduction in any medium,provided you give appropriate credit to the original author(s) and thesource, provide a link to the Creative Commons license, and indicate ifchanges were made.

References

1. Chinese health promotion fundation editorial board of ChineseWhite Papers of osteoporosis (2009) Chinese White Papers of os-teoporosis. Chin J Health Manag 3:148–154

2. Cui L, Chen L, Xia W et al (2017) Vertebral fracture in postmeno-pausal Chinese women: a population-based study. Osteoporos Int28:2583–2590

3. Si L, Winzenberg TM, Jiang Q, Chen M, Palmer AJ (2015)Projection of osteoporosis-related fractures and costs in China:2010-2050. Osteoporos Int 26:1929–1937

4. Kanis JA, Adams J, Borgstrom F, Cooper C, Jonsson B, Preedy D,Selby P, Compston J (2008) The cost-effectiveness of alendronatein the management of osteoporosis. Bone 42:4–15

5. ChineseMedical Association the Branch of Osteoporosis and BoneMineral Disease (2017) Guideline of diagnosis and treatment ofprimary osteoporosis. Chin J Osteoporos Bone Miner Res 10:413–444

6. FRAX (2017) WHO fracture risk assessment tool. . https://www.sheffield.ac.uk/FRAX/tool.aspx?country=2. Accessed 07/10 2017

7. Kanis JA, McCloskey EV, Johansson H, Strom O, Borgstrom F,Oden A (2008) Case finding for the management of osteoporosiswith FRAX–assessment and intervention thresholds for the UK.Osteoporos Int 19:1395–1408

8. Lippuner K, Johansson H, Borgstrom F, Kanis JA, Rizzoli R (2012)Cost-effective intervention thresholds against osteoporotic fracturesbased on FRAX(R) in Switzerland. Osteoporos Int 23:2579–2589

9. Makras P, Athanasakis K, Boubouchairopoulou N, Rizou S,Anastasilakis AD, Kyriopoulos J, Lyritis GP (2015) Cost-effective osteoporosis treatment thresholds in Greece. OsteoporosInt 26:1949–1957

10. Marques A, Lourenco O, Ortsater G, Borgstrom F, Kanis JA, daSilva JA (2016) Cost-effectiveness of intervention thresholds forthe treatment of osteoporosis based on FRAX((R)) in Portugal.Calcif Tissue Int 99:131–141

11. Zhang Z, Ou Y, Sheng Z, Liao E (2014) How to decide interventionthresholds based on FRAX in central south Chinese postmenopaus-al women. Endocrine 45:195–197

12. Cheung E, Cheung CL, Kung AW, Tan KC (2014) Possible FRAX-based intervention thresholds for a cohort of Chinese postmeno-pausal women. Osteoporos Int 25:1017–1023

13. Chang AJ, Ying Q, Chen XN, Wang WM, Chen N (2016)Evaluation of three risk assessment tools in discriminating fracturestatus among Chinese patients undergoing hemodialysis.Osteoporos Int 27:3599–3606

14. Cheung EY, Bow CH, Cheung CL, Soong C, Yeung S, Loong C,Kung A (2012) Discriminative value of FRAX for fracture predic-tion in a cohort of Chinese postmenopausal women. Osteoporos Int23:871–878

15. Zhao J, Xia W, Nie M et al (2012) A haplotype of MATN3 isassociated with vertebral fracture in Chinese postmenopausal wom-en: Peking Vertebral Fracture (PK-VF) study. Bone 50:917–924

16. Li N, Wang X, Jiang Y et al (2014) Association of GALNT3 genepolymorphisms with bone mineral density in Chinese postmeno-pausal women: the Peking Vertebral Fracture study. Menopause 21:515–521

17. Zhao J, Xia W, Nie M et al (2011) The levels of bone turnovermarkers in Chinese postmenopausal women: Peking VertebralFracture study. Menopause 18:1237–1243

18. Bow CH, Cheung E, Cheung CL et al (2012) Ethnic difference ofclinical vertebral fracture risk. Osteoporos Int 23:879–885

19. Wang J, Wang Y, Liu WD, Wang F, Yin ZS (2014) Hip fractures inHefei, China: the Hefei osteoporosis project. J Bone Miner Metab32:206–214

20. Lofthus CM, Frihagen F, Meyer HE, Nordsletten L, Melhuus K,Falch JA (2008) Epidemiology of distal forearm fractures in Oslo,Norway. Osteoporos Int 19:781–786

21. Melton LJ 3rd, Thamer M, Ray NF, Chan JK, Chesnut CH 3rd,Einhorn TA, Johnston CC, Raisz LG, Silverman SL, Siris ES(1997) Fractures attributable to osteoporosis: report from theNational Osteoporosis Foundation. J Bone Miner Res 12:16–23

22. Kanis J, Brazier J, Stevenson M, Calvert N, Lloyd Jones M (2003)Treatment of established osteoporosis: a systematic review andcost-utility analysis. Health Technol Assess 6:146

23. Bliuc D, Nguyen ND, Milch VE, Nguyen TV, Eisman JA, CenterJR (2009) Mortality risk associated with low-trauma osteoporoticfracture and subsequent fracture in men and women. JAMA 301:513–521

24. Sun S, Chen J, JohannessonM,Kind P, Xu L, ZhangY, BurstromK(2011) Population health status in China: EQ-5D results, by age,sex and socio-economic status, from the National Health ServicesSurvey 2008. Qual Life Res 20:309–320

25. Si L, Winzenberg TM, de Graaff B, Palmer AJ (2014) A systematicreview and meta-analysis of utility-based quality of life forosteoporosis-related conditions. Osteoporos Int 25:1987–1997

26. Black DM, Delmas PD, Eastell R et al (2007) Once-yearly zoledro-nic acid for treatment of postmenopausal osteoporosis. N Engl JMed 356:1809–1822

27. Tremblay E, Perreault S, Dorais M (2016) Persistence withdenosumab and zoledronic acid among older women: apopulation-based cohort study. Arch Osteoporos 11:30

28. Pharmcube: Market drug price database of China. http://db.pharmcube.com/database/drugib/index Accessed Feb 07 2019

29. China Center for Health Economic Research (2011) ChinaGuidelines for Pharmacoeconomic Evaluations. Beijing.

30. Cramer JA, Roy A, Burrell A, Fairchild CJ, FuldeoreMJ, OllendorfDA, Wong PK (2008) Medication compliance and persistence: ter-minology and definitions. Value Health 11:44–47

31. Qu B, Ma Y, Yan M, Wu HH, Fan L, Liao DF, Pan XM, Hong Z(2014) The economic burden of fracture patients with osteoporosisin western China. Osteoporos Int 25:1853–1860

32. Inflation: current and historic inflation by country. http://www.inflation.eu. Accessed Feb 07 2019

33. National Bureau of Statistics (2016) Status of deaths by age and sex(2011.11.1-2012.10.30). 2012 National Sample Survey onPopulation Changes. China Statistics Press

34. Chung M, Lee J, Terasawa T, Lau J, Trikalinos TA (2011) VitaminDwith or without calcium supplementation for prevention of cancerand fractures: an updated meta-analysis for the U.S. PreventiveServices Task Force. Ann Intern Med 155:827–838

316 Osteoporos Int (2020) 31:307–316

35. Wells GA, Cranney A, Peterson J, BoucherM, Shea B, RobinsonV,Coyle D, Tugwell P (2008) Alendronate for the primary and sec-ondary prevention of osteoporotic fractures in postmenopausalwomen. Cochrane Database Syst Rev Cd001155

36. Nayak S, Roberts MS, Greenspan SL (2011) Cost-effectiveness ofdifferent screening strategies for osteoporosis in postmenopausalwomen. Ann Intern Med 155:751–761

37. Cheng TT, Yu SF, Hsu CY, Chen SH, Su BY, Yang TS (2013)Differences in adherence to osteoporosis regimens: a 2-year analy-sis of a population treated under specific guidelines. Clin Ther 35:1005–1015

38. Cheng XG, Yang DZ, Zhou Q et al (2007) Age-related bone min-eral density, bone loss rate, prevalence of osteoporosis, and refer-ence database of women at multiple centers in China. J ClinDensitom 10:276–284

39. Park C, Ha YC, Jang S, Jang S, Yoon HK, Lee YK (2011) Theincidence and residual lifetime risk of osteoporosis-related fracturesin Korea. J Bone Miner Metab 29:744–751

40. WHO (2019) Life expectancy: life tables by country China. WorldHealth Organization. . http://apps.who.int/gho/data/view.main.60340?lang=en. Accessed 06-24 2019

41. Chinese medical associaton branch of osteoporosis and bone min-eral disease (2017) Guideline of prevention and treatment of prima-ry osteoporosis. Chin J Osteoporos Bone Miner Res 10:413–443

42. Jiang Y, Ni W (2016) Expected lifetime numbers, risks, and burdenof osteoporotic fractures for 50-year old Chinese women: a discreteevent simulation incorporating FRAX. J Bone Miner Metab 34:714–722

43. Ni W, Jiang Y (2017) Evaluation on the cost-effective threshold ofosteoporosis treatment on elderly women in China using discreteevent simulation model. Osteoporos Int 28:529–538

44. Si L, Winzenberg TM, Chen M, Jiang Q, Neil A, Palmer AJ (2016)Screening for osteoporosis in Chinese post-menopausal women: ahealth economic modelling study. Osteoporos Int 27:2259–2269

45. Si L,Winzenberg TM, JiangQ, Palmer AJ (2015) Screening for andtreatment of osteoporosis: construction and validation of a state-transition microsimulation cost-effectiveness model. OsteoporosInt 26:1477–1489

46. Tosteson AN, Melton LJ 3rd, Dawson-Hughes B, Baim S, FavusMJ, Khosla S, Lindsay RL (2008) Cost-effective osteoporosis treat-ment thresholds: the United States perspective. Osteoporos Int 19:437–447

47. Liberman UA, Weiss SR, Broll J, Minne HW, Quan H, Bell NH,Rodriguez-Portales J, Downs RW Jr, Dequeker J, Favus M (1995)Effect of oral alendronate on bonemineral density and the incidenceof fractures in postmenopausal osteoporosis. The AlendronatePhase III Osteoporosis Treatment Study Group. N Engl J Med333:1437–1443

48. Liu GG, Wu H, Li M, Gao C, Luo N (2014) Chinese time trade-offvalues for EQ-5D health states. Value Health 17:597–604

49. Hiligsmann M, Reginster JY, Tosteson ANA et al (2019)Recommendations for the conduct of economic evaluations in os-teoporosis: outcomes of an experts’ consensus meeting organizedby the European Society for Clinical and Economic Aspects ofOsteoporosis, Osteoarthritis and Musculoskeletal Diseases(ESCEO) and the US branch of the International OsteoporosisFoundation. Osteoporos Int 30:45–57

50. Husereau D, Drummond M, Petrou S, Carswell C, Moher D,Greenberg D, Augustovski F, Briggs AH, Mauskopf J, Loder E(2013) Consolidated Health Economic Evaluation ReportingStandards (CHEERS) statement. Pharmacoeconomics 31:361–367

Publisher’s note Springer Nature remains neutral with regard to jurisdic-tional claims in published maps and institutional affiliations.